INDEX

Explanations

sequences of numerical values

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

Data Type

torch.float32

Hook Point

blocks.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

xual

-0.86

 mutated

-0.75

 brid

-0.73

 habit

-0.73

 spiders

-0.73

 describ

-0.71

 claws

-0.70

 destro

-0.69

 garments

-0.69

 ranged

-0.69

POSITIVE LOGITS

October

1.42

January

1.42

December

1.41

November

1.38

September

1.36

August

1.33

July

1.33

February

1.32

June

1.29

April

1.27

Activations Density 0.045%

sequences of numerical values

No Comments

No Known Activations